1771
The Journal of Experimental Biology 201, 1771–1784 (1998)
Printed in Great Britain © The Company of Biologists Limited 1998
JEB1395
TAIL-FLIP MECHANISM AND SIZE-DEPENDENT KINEMATICS OF ESCAPE
SWIMMING IN THE BROWN SHRIMP CRANGON CRANGON
STEPHEN A. ARNOTT1,2,*, DOUGLAS M. NEIL1 AND ALAN D. ANSELL2
of Biomedical and Life Sciences, Graham Kerr Building, University of Glasgow, Glasgow G12 8QQ,
Scotland, UK and 2Dunstaffnage Marine Laboratory, PO Box 3, Oban, Argyll PA34 4AD, Scotland, UK
1Institute
*e-mail: [email protected]
Accepted 9 March; published on WWW 12 May 1998
Summary
Tail-flip escape swimming by the brown shrimp Crangon
The kinematic properties of tail-flips were found to vary
crangon has been investigated across a range of body
with shrimp size. As shrimp body length increased, the rate
lengths (11–69 mm) using high-speed video analysis. This
of body flexion and re-extension decreased whilst the
duration of tail-flips increased. Mean (and maximum)
has revealed several novel aspects of the tail-flip
velocity estimates ranged between 0.4 m s−1 (0.7 m s−1) and
mechanism when compared with that of other decapod
crustaceans that have been studied. (i) The pattern of body
1.1 m s−1 (1.8 m s−1) for shrimps of different sizes. The
flexion in C. crangon produces movement of the
combined effects of escape behaviour and size-dependent
cephalothorax as well as the abdomen about the centre of
variability in tail-flip kinematics will have important
mass. (ii) Shrimps form a ‘head-fan’ with their antennal
implications with regard to predation risk.
scales, in addition to the tail-fan formed by their uropods,
apparently for generating thrust during tail-flips. (iii)
Shrimps typically swim on their side rather than in an
upright body position. It is suggested that these features
Key words: Crangon crangon, escape, caridean shrimp, giant fibres,
size, swimming, tail-flip, kinematics.
may be interlinked and derive from habitat specialisation.
Introduction
Many malacostracan crustaceans such as crayfish, lobsters
and shrimps possess an elongated abdomen that can be used
for propelling the animal through the water with a powerful
tail-flip swimming action. This energetically expensive
behaviour is primarily employed as a startle escape response
when they are attacked by a predator, but may also occur in
response to noxious substances, during feeding (Wine and
Krasne, 1972; Bellman and Krasne, 1983) and during agonistic
encounters (Edwards, 1995).
The propulsive forces generated during tail-flip swimming
derive from a combination of reactive forces (added mass) and
resistive forces (drag), with the first of these dominating
instantaneous thrust, and a major contribution being made by
the expanded uropods (Webb, 1979; Neil and Ansell, 1995). A
hydrodynamic ‘squeeze’ force is also produced towards the
end of the tail-flip as the abdomen and cephalothorax are
brought towards one another (Daniel and Meyhöfer, 1989).
Since the uropods generate a significant proportion of the
thrust, the tail-flip of many decapods represents a ‘single oar’
rowing action that produces a combination of both translatory
forces (displacing the centre of mass) and rotational forces
(pitching the animal forwards). Calculations by Daniel and
Meyhöfer (1989) indicate that total tail-flip force and the
balance between translatory and rotational forces scale
differently from one another with respect to animal size.
Therefore, for an animal of given dimensions and with an
isometric growth pattern, there is a particular size at which tailflip performance will be optimal.
The neuronal control of tail-flip behaviour has been
intensively studied, particularly in the crayfish Procambarus
clarkii. In this species, both giant and non-giant neuronal
networks may mediate the tail-flips. There are two pairs of
giant fibres, the lateral giants (LGs) and the medial giants
(MGs), that produce upward and posteriorly directed escapes,
respectively (Wine, 1984; Krasne and Wine, 1988), and this
pattern of behaviour is typical of several other decapod species
(e.g. Webb, 1979; Jacklyn and Ritz, 1986; Newland and Neil,
1990a).
Among the crustaceans, a vast range of body morphologies,
body sizes, habitat types and life styles exist. Adaptations of
the tail-flip mechanism are required to accommodate these
divergences. For example, at the neuronal level, whilst groups
such as crayfish (Astacidea) and caridean shrimps (Caridea)
possess two pairs of giant fibres (Johnson, 1924), mud shrimps
(Thalassinoidea) possess just one pair (the LGs) and squat
lobsters (Galatheidae) possess neither (Paul, 1990).
Intraspecific differences also occur, as demonstrated at the
behavioural level in the American lobster Homarus
1772 S. A. ARNOTT, D. M. NEIL AND A. D. ANSELL
americanus. Juveniles of this species have a comparatively
large abdomen and small claws and respond to predators by
tail-flipping, whereas adults have a comparatively small
abdomen and large claws, and respond to predators with
defensive displays (Lang et al. 1977).
In this investigation, we have examined tail-flip swimming
in the brown shrimp Crangon crangon, an epibenthic caridean
shrimp that is widespread in shallow soft-bottom bays and
estuaries around Europe. It lives on or buried just beneath the
sediment surface (Al-Adhub and Naylor, 1975; Pinn and
Ansell, 1993), and ranges in total length between 5 and 90 mm,
although maximum lengths of approximately 70 mm are more
typical (Tiews, 1970). The species is heavily fished in some
areas and forms an important prey item for a number of
commercially important fish species (Tiews, 1970). As part of
their defence against both predators (Tallmark and Evans,
1986) and approaching trawl gear (Berghahn et al. 1995),
brown shrimps have a rapid tail-flip escape response. We have
used high-speed video techniques (i) to examine the tail-flip
mechanism of brown shrimps, and to compare and contrast this
with respect to habitat type and with respect to other crustacean
species, and (ii) to quantify the kinematic performance of tailflip swimming with respect to body length. Our results confirm
that size has a significant effect upon escape performance and
reveal several novel aspects of tail-flip swimming and body
orientation in C. crangon that distinguish their behaviour from
that of larger, more heavily calcified crustaceans. These
differences suggest that there is an intrinsic relationship
between habitat type, tail-flip mechanism and body orientation
that has important implications with regard to escape from
predatory attacks.
Materials and methods
Animals
Brown shrimps Crangon crangon (L.) were caught during
July 1993 and 1994, in a hand-held trawl net at a depth of less
than 1 m in Dunstaffnage Bay on the west coast of Scotland
and transferred to aquaria (100 cm×50 cm×30 cm with a 1–2 cm
sand substratum) maintained at an approximate salinity of
30 ‰ and temperature of 13 °C (fluctuating with ambient sea
conditions). For predator-evasion experiments, juvenile cod
(Gadus morhua) were used as predators. O-group cod of
between 61–107 mm (tip of snout to tip of caudal fin) were
caught in July 1993 in Dunstaffnage Bay (<2 m depth) using a
beach seine net. These fish were housed in 1 m diameter tanks
with circulating sea water of the same salinity and temperature
as that of the shrimps.
The shrimps were kept for approximately 2 weeks before
being used in experiments, and were fed ad libitum every
second day on chopped mussels and/or mysids. None of the
experimental shrimps was in a berried condition (i.e. carrying
eggs attached to its pleopods), and all of them had a hard
exoskeleton and showed no obvious signs of poor health or
damage. Total body lengths were determined by measuring the
distance from the anterior tip of the rostrum to the posterior tip
of the telson. Twenty-five shrimps of between 11 and 69 mm
body length (wet mass 0.013–4.55 g) were used for high-speed
video experiments. An additional 38 shrimps of between 6 and
36 mm body length (0.002–0.55 g) were used for predatorevasion experiments.
Morphometrics
Total body length (L) was measured to the nearest 0.1 mm
as described above, except for the relationship between L and
the abdomen length (A). In this case, L was measured to the
base of the rostrum rather than to its anterior tip. Abdomen
length was measured as the distance from the posterior edge
of the carapace to the posterior tip of the telson along the
midline of the shrimp’s dorsal side.
The cross-sectional area of the abdominal muscle was
determined in shrimps that had been preserved in a flat position
(i.e. abdomen extended) for 24 h in formaldehyde solution (4 %
v/v in sea water). Total length was measured both before and
after fixing, but no shrinkage effects were detected as a result
of the procedure. Transverse sections were cut half-way along
the third abdominal segment and examined under a microscope
(magnification ×6–25) linked to a video recording system
(Kappa CF 11/2 camera, Panasonic NV-FS200 HQ VHS
recorder). The cross-sectional area of the fast muscles (F, the
flexor and extensor areas combined) was measured from
digitised video images using NIH Image 1.55 computer
software. The area attributable to superficial pleopod and slow
muscles was not included in the analysis.
Wet mass of live shrimps (blotted dry) was measured on a
balance to the nearest milligram.
Protocol for high-speed video experiments
All experiments were conducted in an experimental arena
(diameter 1 m, seawater depth 17 cm) in an air-conditioned
room at 13 °C and were filmed from directly above using a
NAC high-speed video camera linked to a NAC HSV400 video
recorder. This provided a view of the horizontal position of the
shrimp within the arena (‘camera view’). A mirror was placed
on the bottom of the arena at 45 ° to the camera to provide a
view of the shrimp’s vertical elevation above the substratum
(‘mirror view’). A 5 or 10 cm marker on the bottom of the arena
provided calibration marks on the video films. Illumination
was provided by a synchronised strobe, and the light from this
was orientated along the axis of the camera lens by reflecting
it in a half-silvered mirror angled just in front of the camera
lens. The base of the arena was covered with reflective material
(3M Scotchlite) to produce a sharp silhouette image of the
shrimp in the camera view. A silhouette image was also
obtained in the mirror view by placing an upright board
covered in 3M Scotchlite at the opposite end of the arena from
the mirror. All experiments were recorded at 200 frames s−1 on
the high-speed video recorder.
For each experiment, a shrimp was removed from its holding
tank by pressing lightly down on its carapace and then lifting
it between two fingers. This method tended to inhibit the tailflip escape response (a similar response has been noted in
Tail-flip swimming in brown shrimps 1773
crayfish; see Krasne and Wine, 1975) and therefore enabled
shrimps to be moved without inducing muscle fatigue. The
shrimp was placed on the bottom of the experimental arena and
covered for 10 min with an upturned transparent plastic
container in which perforations had been made. During this
period, the water was aerated with an air-stone. At the start of
an experiment, the video recording equipment was turned on
and the plastic container and air-stone were removed. Tail-flip
escape responses were induced, either by a rapid flick with a
submerged finger placed 5–10 cm from the shrimp or by
rapidly propelling a partially submerged rod (2 cm diameter)
towards the shrimp. No direct physical contact was made with
the shrimp or substratum; the stimulus therefore comprised
mainly visual and water-borne vibrational cues. Experiments
on dead animals confirmed that no passive movement of the
shrimp was created by water displacement arising from either
of the stimuli. Each shrimp was made to perform between 1
and 5 multiple tail-flip swimming bouts, during which no
obvious signs of physical exhaustion were visible.
Protocol for predator-evasion experiments
An additional set of experiments was conducted using 38
shrimps (6–36 mm) and 38 cod (61–107 mm). The experiments
were conducted at 13 °C in a circular arena with a white
reflective substratum, a diameter of 30 cm and a water depth
of 20 cm. Predatory encounters between cod and shrimps were
filmed with a camera placed directly above the arena, and
recorded on a conventional video recorder (Panasonic AG6024) at a frame rate of 50 frames s−1. A time inserter (IMP
Electronics V9000) was used for inserting elapsed time (0.01 s)
onto the video recordings.
For each experiment, a single shrimp and cod were placed
in the arena and allowed to settle for 15 min. An upturned
opaque container with perforations in it was placed over the
shrimp during this period, and aeration was provided by means
of an air stone. At the start of the experiment, the air was turned
off and the container was lifted remotely from behind a screen
using an attached string. Experiments were filmed for 1 h or
until the shrimp had been consumed by the cod.
Estimation of the centre of mass
The centre of mass in Crangon crangon was determined by
suspending frozen specimens between two opposed points
formed by fine pins mounted on a pair of forceps. Shrimps
were frozen (−10 °C) with their abdomen fully extended (i.e.
in their normal resting body posture) or with their abdomen
fully flexed in order to determine the shift in position of the
centre of mass during the course of a tail-flip. In each case, the
position of the pins on the shrimp was adjusted in air until the
animal could be placed in any pitch orientation without
pivoting under its own weight. The centre of mass was then
assumed to lie on the axis between the pin attachment points.
Using this method, the centre of mass was found to lie within
the ventral half of the first abdominal segment when the shrimp
was in a fully extended position. When the body was fully
flexed, the centre of mass shifted slightly to a position level
with the coxa of the fifth pereiopod. Therefore, a single
intermediate point on the postero-ventral corner of the shrimp’s
cephalothorax was used for digitising the estimated centre of
mass from video film (see below).
Analysis of high-speed video recordings
Only escapes in which the shrimp performed more than one
tail-flip during an escape swimming bout were analysed (this
was the typical response to the type of stimulus used). Of 89
multiple tail-flip swimming bouts that were filmed, 25 were
selected for kinematic analysis on the basis that the shrimp was
swimming above and parallel to the substratum (identified
from the mirror view of the shrimp). High-speed video
sequences were replayed frame by frame onto a monitor (JVC)
from which reference points on the shrimp’s body were
digitised on an attached digitising tablet (NAC). These data
were analysed using MOVIAS 3.00-4 (NAC) and Excel 5.0
(Microsoft) software.
Movement in the horizontal plane was analysed by digitising
four points from the camera view of the shrimp’s lateral aspect
(Fig. 1A; note that shrimps usually swam on their side; see
Results section). These were: 1, the eyes; 2, the leading edge of
the abdomen at its mid-point of flexion; 3, the posterior tip of
the sixth abdominal segment; and 4, the estimated centre of
mass. Points 1, 3 and 4 could be accurately identified by the fact
that they occur at angled joints between the antennal scales and
the cephalothorax (point 1), the sixth abdominal segment and the
telson (point 3), and the cephalothorax and the abdomen (point
4). Although these angled joints were not visible in all images
throughout a tail-flip cycle, once they had been identified for a
particular shrimp, points in adjacent frames could be located by
means of their respective distances from the anterior or posterior
tips of the shrimp. In the first 1–3 frames of an escape swimming
sequence, the camera view of the shrimp comprised images of
the dorsal (or partial-dorsal) aspect as it performed a lateral
rolling manoeuvre (described below). In such instances, the
points were digitised along the midline of the shrimp (or
estimated midline for partial-dorsal aspects). Escape sequences
were analysed from the frame immediately preceding the
shrimp’s initial movement until the frame in which one of the
digitising points on the shrimp’s body moved out of the camera’s
field of view. The 25 sequences analysed consisted of two (N=9),
three (N=9), four (N=5) or five (N=2) tail-flips, but in some
cases, these only included the flexion phase of the last tail-flip.
The body angle of the shrimp was defined as the angle
subtended by points 1, 2 and 3 (Fig. 1A). Changes in this angle
(∆ angle) between successive frames were used to determine
maximum angular velocity and maximum angular acceleration
during flexion or re-extension phases of the tail-flip (i.e. the
peak value attained over a 5 ms interval). The mean angular
velocity for a complete flexion or re-extension movement of
the abdomen was calculated as: (total ∆ angle)/(total time
taken). Negative angular values were assigned to flexion
movements and positive angular values to body re-extension
movements.
Displacement of the shrimp was determined by measuring
1774 S. A. ARNOTT, D. M. NEIL AND A. D. ANSELL
Fig. 1. High-speed video analysis. (A) Lateral
aspect (camera view) of an escaping shrimp
C
A
B
3
showing the four points that were digitised. 1,
2
P2
eyes; 2, leading edge of the abdomen at its midTF
HF
3
point of flexion; 3, posterior tip of the sixth
1
abdominal segment; 4, centre of mass. The
2
4
P1
1
body angle of the shrimp (β) was measured as
4
the angle subtended by points 1, 2 and 3.
β
(B) Digitised points (crosses) of the shrimp’s
centre
of
mass
showing
horizontal
displacement during tail-flips 1–3 of an escape. Stick diagrams join points 1 (uppermost), 2 and 3 (lowermost) at two positions during tail-flip
1. The pitch angles (P1, P2) between successive tail-flips were measured as the angle subtended by the fitted lines, with positive values assigned
to pitch in the rostral direction (as in this example). (C) Dorsal aspect (mirror view) of an escaping shrimp showing measurement of the headfan (HF) and tail-fan (TF) width.
the distance travelled by the digitised positions of the estimated
centre of mass between one frame and the next. Distance
travelled per tail-flip was calculated as the cumulative sum of
these values over a complete body flexion/re-extension cycle.
The mean velocity during an entire multiple tail-flip swimming
bout was calculated as: (cumulative distance travelled)/(time
elapsed). Velocity over 5 ms time intervals was determined
from the distance moved by the centre of mass between one
image and the next. Maximum velocity values for each tail-flip
in an escape swimming bout were derived from the peak values
reached (over a 5 ms interval) during body flexion.
Acceleration values have not been included because of the
increased error associated with calculating such second-order
differentials (see Harper and Blake, 1989).
Rotation in the shrimp’s pitch plane between successive tailflips was estimated by fitting a straight line through the
horizontal trajectory of the centre of mass for each tail-flip. The
angle between successive tail-flip trajectories was measured as
the pitch angle, with positive angles assigned to rotation in the
rostral direction and negative angles to rotation in the caudal
direction (Fig. 1B).
The shrimp’s antennal scales (scaphocerites) and uropods
pivoted laterally during tail-flips, expanding during body
flexion to form a head-fan and tail-fan respectively. These
movements were analysed in a selection of sequences by
digitising the most lateral point of each antennal scale or
uropod (as seen from the shrimp’s dorso-ventral aspect) and
measuring the linear distance between the opposite points (Fig.
1C).
During tail-flips, thrust is produced when appendages are
moved through the water with respect to the centre of mass
(Webb, 1979). In 12 sequences, the positions of points 1 and
3 with respect to point 4 (centre of mass) were determined at
the beginning and end of each flexion phase. From these data,
the distance moved during body flexion by the head-fan and
tail-fan with respect to the centre of mass was calculated (note
that points 1 and 3 are located at the base of the head-fan and
tail-fan respectively).
Analysis of predator-evasion recordings
Predator-evasion experiments were used to provide
supplementary mean velocity measurements for shrimps
responding to a natural predator. Recordings were analysed
from sequences in which shrimps performed multiple tail-flip
escapes in response to an approach by a cod. The estimated
centre of mass of the shrimp was traced frame by frame from
a video monitor (JVC) onto an acetate overlay. These x,y
coordinates were digitised, and distances between successive
points were calculated. Mean horizontal velocity was
calculated as a function of cumulative displacement divided by
time elapsed.
Statistical analyses of data
Statistical calculations were performed using Minitab
10Xtra (Minitab Inc.) software or according to Zar (1996).
Muscle area and wet mass relationships were examined using
linear regression of loge-transformed data. Deviation of
regression slope coefficients from values expected for
isometric growth (2.0 for muscle area, 3.0 for wet mass) was
investigated using t-tests.
Comparisons between first tail-flip versus second tail-flip
measurements for an escape swimming bout were conducted
using two-tailed paired t-tests (where percentages were
compared, data were arcsine-transformed). Relationships
between shrimp body length and various tail-flip
measurements were determined using regression models.
Linear regressions were fitted to data on the duration of tailflip phases (duration of entire tail-flips and of body flexion and
re-extension phases). For data on displacement per tail-flip,
mean velocity and maximum velocity, three types of regression
models were tested. These were (i) a linear regression, (ii) a
linear regression of the log-transformed data, and (iii) a
quadratic regression. The most appropriate model was then
determined for each set of data by plotting the residuals of the
regression values and assessing their deviation from zero. In
all instances, the quadratic model produced the best fit, and
therefore only these regressions are presented.
Results
Morphometrics
The ratio of abdomen length to total length (A:L) remained
stable at approximately 0.77±0.02 (mean ± S.D., N=503) for
shrimps of all lengths.
Tail-flip swimming in brown shrimps 1775
The cross-sectional area of the fast muscle within the third
abdominal segment was only measured in shrimps between 20
and 50 mm total length. Within this range, fast muscle area (F,
mm2) was related to total length (L, mm) by:
F = 0.009L1.987 (N=16, r2=0.97, P<0.0001) .
(1)
The scaling factor of 1.987 was not significantly different from
2.0 (t-test, t15=0.144, P>0.50).
The relationship between shrimp total length and shrimp wet
mass (M, g) was described by the function:
M = 5.52×10−6L3.217 (N=90, r2=0.996, P<0.0001) .
(2)
In this instance, the scaling factor was significantly greater than
3.0 (t-test, t89=9.65, P<0.001).
Description of tail-flips
Flexion and re-extension movements
Shrimps responded to a vibrational or visual stimulus by
performing either a single tail-flip, comprising a single cycle
of abdominal flexion and re-extension, or more typically, a
series of multiple tail-flips (termed an escape swimming bout).
Escape latencies (the time between the first visible movement
of the stimulus until the first visible movement of the shrimp)
were between 10 and 15 ms.
The kinematic variables of an escape swimming bout
performed by an 11 mm total length shrimp are shown in Fig. 2
and are analysed in detail below in relation to body length and
Body angle
(degrees)
A
180
135
90
45
0
B
Velocity (m s−1)
0
20
40
60
80
100
Time (ms)
1
120
140
C
0.8
0.6
0.4
0.2
0
0
20
40
60
80
100
Time (ms)
120
140
Fig. 2. (A) Sequence of superimposed stick diagrams digitised from
high-speed video (camera view; points 1, 2 and 3 of an 11 mm
shrimp) during the first three tail-flips of an escape swimming bout.
Point 1 (eyes) is uppermost, and the direction of travel is from left to
right. Graphs show the corresponding changes in body angle (B) and
velocity of the centre of mass (C). Filled symbols represent the
beginning of each flexion phase.
tail-flip number. Broadly speaking, the flexion and reextension movements were similar between the first and
subsequent tail-flips within a swimming bout and for shrimps
of all sizes. They resulted in both point 1 (the eyes) and point
3 (the telson) pivoting with respect to point 4 (the centre of
mass) during the tail-flip cycle, producing a ‘jack-knife’
pattern of movement. The pivoting movements of point 1 were
not as great as those of point 3, as shown by an analysis of 12
of the sequences (shrimp total body lengths 33–69 mm).
Expressed as a percentage, the distance (with respect to point
4) travelled by point 1 divided by the distance travelled by
point 3 ranged between 33 and 91 % (mean 55 %) during the
tail-flip 1 flexion and between 29 and 88 % (mean 44 %) during
the tail-flip 2 flexion. The difference between tail-flip 1 and 2
was not significant (paired t-test, t11=1.40, P=0.19).
The flexion phase of the tail-flip (see 0–30 ms of Fig. 3) was
brought about predominantly by movements within the
anterior portion of the abdomen, with the mid-point for flexion
being located at a point between abdominal segments 2 and 3.
Virtually no movement occurred at the joint between the sixth
abdominal segment and the telson. During the re-extension
phase of the tail-flip (see 40–110 ms of Fig. 3), the anterior
abdominal segments were extended earlier than the posterior
segments, and the joint between segment 6 and the telson was
held in a flexed position until the next flexion phase was
initiated. Movement about this latter joint was probably
brought about (at least in part) by passive forces exerted by the
incident flow of water.
Articulation of the uropods and antennal scales
During the flexion phase of the tail-flip, both the antennal
scales and the uropods underwent lateral pivoting movements
to form expanded propulsive surfaces (the head-fan and tailfan surfaces respectively). In a 31 mm total length shrimp, both
fans had a maximum width of approximately 10 mm (Fig. 4).
Full expansion of the tail-fan occurred within 5–10 ms of the
start of flexion, and within 10–15 ms for the head-fan, but each
fan remained maximally spread for less than 5 ms (the duration
of a single frame). Towards the end of the flexion phase, both
fans were gradually retracted so that, by the start of body reextension, the width of the tail-fan (as seen from the dorsoventral aspect) was only 15–20 % of its maximum, whilst that
of the head-fan was between 40 and 60 % of its maximum. The
difference between the minimum widths is due to the fact that
shrimps not only retract the uropods laterally but also fold the
opposing ventral surfaces together beneath the telson into a
streamlined position. Reduction in the width of the head-fan
occurred primarily as a result of pivoting the antennal scales
medially rather than by folding them.
Orientation of the body during tail-flip swimming
In the majority of escapes filmed (84 %), the first tail-flip of
a swimming bout was accompanied by a lateral roll of the
shrimp’s body about its antero-posterior axis so that the animal
escaped either to its left or right side (Fig. 5). This rotation was
evident within the first 1–3 frames in which movement was
1776 S. A. ARNOTT, D. M. NEIL AND A. D. ANSELL
10
0
30
20
40
50
HF
Fig. 3. High-speed video images (every
second frame shown) of tail-flip 2 in an
escape swimming bout. The shrimp is
viewed from its lateral aspect with the headfan (HF) to the left and tail-fan (TF) to the
right at time zero. Numbers refer to the time
elapsed (in milliseconds) since the first
frame. 0–30 ms, flexion phase; 40–110 ms,
re-extension phase.
TF
60
detected (i.e. within 5–15 ms of the onset of movement) and
caused subsequent tail-flips to take place with the shrimp
swimming on its side in a horizontal direction.
In some responses (16 %), lateral roll was much less
pronounced or absent, and the first flexion of an escape
response was instead performed with the shrimp orientated in
an upright position, producing a predominantly vertical
trajectory. In these cases, a roll was usually executed during
the first re-extension (typically accompanied by pleopod
movements), so that subsequent tail-flips of the escape then
also took place with the shrimp swimming on its side in a
horizontal direction. These mainly vertical escapes occurred in
response to both rostrally and caudally applied stimuli.
In three instances (3 %) in which an escape involved an
initial vertical rather than lateral displacement, no roll occurred
at all, even during the first re-extension. Instead, the shrimp
rotated rostrally (i.e. pitched forward) during the first reextension and continued doing so during the second flexion,
causing it to perform a partial forward somersault. Subsequent
tail-flips then took place above the substratum in a
predominantly horizontal direction but, instead of the shrimp
swimming on its side, it swam with its cephalothorax
positioned lowermost and its abdomen uppermost.
Following a sideways roll during the first tail-flip,
subsequent tail-flips often involved a smaller degree of roll
70
80
90
100
110
which tended to direct the shrimp slightly downwards rather
than maintaining a constant vertical elevation. This looping
movement resulted in periodic contact with the substratum
which, on a sand or mud bottom, caused intermittent puffs of
sediment to be produced. Similar observations were made by
Tallmark and Evans (1986).
Pitch movements during subsequent tail-flips of an escape
response
When a shrimp was swimming on its side (the typical escape
mode observed), horizontal displacement during each flexion
phase occurred along an approximately linear trajectory.
Horizontal steering was achieved by rotation in the pitch plane
between one tail-flip and the next, thereby adjusting the
direction of travel. When changes in direction did take place,
the largest adjustments tended to be performed during the first
few tail-flips of an escape, after which steering adjustments
were minimal. Larger changes in direction were observed
when pitching rostrally (up to 70–80 °) than when pitching
caudally (up to 10–15 °).
During the re-extension phase immediately preceding a
large rotational pitch in the rostral direction, a single beat of
the pleopods sometimes occurred, possibly assisting in
bringing about the change of direction (this can be seen in
Fig. 5).
Fig. 4. High-speed video images
TF
25
30
35
0
20
showing expansion of the head-fan (HF)
and tail-fan (TF) during an escape
swimming bout. Top row: camera view
of the shrimp (view from vertically
HF
above the arena). Bottom row:
concurrent mirror view (looking
25
30
35
0
20
horizontally) showing the shrimp
approximately
5 cm
above
the
substratum (S). The initial pair of
images (left-hand column) show the reTF
HF
extension phase of tail-flip 1, during
S
which the head-fan and tail-fan are
retracted into a streamlined position.
During the subsequent flexion phase of tail-flip 2 (columns 2–5), the head-fan and tail-fan are rapidly expanded. Numbers refer to time elapsed
(in milliseconds).
Tail-flip swimming in brown shrimps 1777
0
HF
Fig. 5. High-speed video images from the camera
view of the shrimp (view from vertically above the
arena, every second frame) showing laterally directed
body roll during the first tail-flip of an escape
response. Numbers refer to time elapsed (in
milliseconds). The shrimp starts at time zero from an
upright position, stationary on the substratum. HF,
head-fan; TF, tail-fan. In this example, movement of
the pleopods (P) can also be seen during the reextension phase (80–100 ms).
10
30
20
50
40
TF
Movement in the shrimp’s yaw plane
Rotation of the shrimp in the yaw plane was not examined
in detail, but this does occur and adds to the complexity of tailflips. Yaw rotation was especially evident during the reextension phase of tail-flips. In many of the escapes, when
shrimps were swimming on their side, yaw rotation resulted in
the shrimp’s longitudinal axis being at an angle (rather than
parallel) to the horizontal, with the tail-fan elevated a greater
distance above the substratum than the head-fan.
60
70
100
fully flexed position was 25.0±4.9 ° (mean ± S.D., N=62). There
was no significant difference between the minimum body angle
at the end of the first and second flexions of an escape (paired
t-test, t24=1.68, P=0.11).
The maximum body angle attained at the end of the reextension phase of a tail-flip was more variable (range
75–165 °). The mean angle of all pooled data (tail-flips 1–5)
was 128.2±20.3 ° (mean ± S.D., N=60), but a significant
160
Duration (ms)
A
120
80
40
0
0
Duration (ms)
50
10
20
30
40
50
60
70
B
25
0
0
10
20
30
40
50
60
70
100
Duration (ms)
Body angle measurements
Consecutive flexion and re-extension movements of the
abdomen during a tail-flip swimming bout resulted in cyclic
changes in the body angle (Fig. 2B). Tail-flips usually resulted
in full flexion of the abdomen, causing the tail-fan to come into
close or direct contact with the cephalothorax. Regression
analysis on the pooled data showed that there was no
significant change in minimum body angle with total body
length (t-test on slope of line, t49=1.52, P=0.14). For all tailflips analysed (tail-flips 1–5), the mean body angle when in a
90
P
Tail-flip kinematics
Duration of tail-flip
Tail-flips of the smallest shrimps (11 mm total length) had
typical durations of between 30 and 50 ms, whereas those of
larger shrimps (55–69 mm) were between 65 and 140 ms. Total
tail-flip duration, flexion duration and re-extension duration all
increased as positive linear functions of shrimp body length
(Fig. 6A–C; Table 1). The ratios of flexion duration to total
tail-flip duration for tail-flips 1 and 2 had values of 0.47±0.10
and 0.39±0.08 (mean ± S.D.), respectively, and did not change
significantly with body length (t-tests on regression slopes;
t24=1.83 and t19=0.53 respectively, both P values >0.05).
Within each escape swimming bout, there was no significant
difference between the total duration of tail-flip 1 and that of
tail-flip 2 (paired t-test, t19=1.56, P=0.14). The flexion phase
of tail-flip 2 had a significantly shorter duration than that of
tail-flip 1 (paired t-test, t24=2.70, P=0.012), probably in part
because the body started from a fully extended position at the
beginning of tail-flip 1 compared with a partially extended
position in subsequent tail-flips. However, the re-extension
phase of the tail-flip 2 had a significantly longer duration than
that of tail-flip 1 (paired t-test, t19=2.44, P=0.025).
80
C
75
50
25
0
0
10
20 30
40 50
Body length (mm)
60
70
Fig. 6. Relationships between shrimp total body length and the
duration of tail-flip phases. (A) Duration of an entire tail-flip (flexion
+ re-extension). (B) Duration of the flexion phase. (C) Duration of
the re-extension phase. Filled symbols represent tail-flip 1; open
symbols represent tail-flip 2. Regressions (see Table 1) are either for
the pooled data (A) or only for the first tail-flip data (B,C),
depending upon the results of paired t-tests (see text).
1778 S. A. ARNOTT, D. M. NEIL AND A. D. ANSELL
Table 1. Regression parameters derived for kinematic relationships with shrimp total length
Regression model
Parameter
Data used for fitting regression
Linear regressions
Duration (ms)
Entire tail-flip (tail-flips 1 and 2)
Flexion 1
Flexion 2
Re-extension 1
Re-extension 2
Maximum body angle (degrees)
End of re-extensions 1–5
Mean angular velocity (degrees s−1)
Flexion 1
Flexion 2
Re-extension 1
Re-extension 2
Maximum angular velocity (degrees s−1)
Flexion 1
Flexion 2
Re-extensions 1 and 2
Maximum angular acceleration (degrees s−2)
Flexions 1 and 2
Quadratic regressions
Displacement per tail-flip (mm)
Tail-flips 1–5
Mean velocity (m s−1)
Entire escape swimming bout
Maximum velocity (m s−1)
Flexions 1–5
Regression
ANOVA
c
P
r2
Figure
27.0
16.9
12.5
6.23
22.6
<0.001
<0.001
<0.001
<0.001
<0.01
0.57
0.56
0.53
0.49
0.27
6A
6B
−
6C
−
113
<0.004
0.12
−
70.9
29.0
−37.4
−33.6
−7777
−4639
4360
3833
<0.001
<0.002
<0.001
<0.001
0.61
0.34
0.37
0.32
7A
7A
7A
7A
93.1
70.9
−55.8
−12 336
−8929
7106
<0.001
<0.001
<0.009
0.52
0.35
0.32
7B
7B
7B
−228 573
<0.001
0.57
7C
<0.001
0.60
8A
a
b
1.11
0.34
0.40
0.82
0.56
0.41
2486
2.88
−0.02
−14.6
0.0415
−0.0004
−0.008
<0.001
0.77
8B
0.0596
−0.00051
0.082
<0.001
0.58
8C
L, total body length (mm).
For linear regressions, y=aL+c; for quadratic regressions, y=aL+bL2+c.
Figure refers to the figure number in which the regression is shown fitted to the data.
increase with shrimp total body length was detected (Table 1).
The predicted values from the regression indicated that the
mean angle increased from 118 to 141 ° with an increase in
body length from 11 to 69 mm, but this relationship accounted
for very little of the overall variability (r2=0.12). There was no
significant difference between the maximum body angles at the
end of re-extensions 1 and 2 (paired t-test, t19=1.51, P=0.14).
Mean and maximum angular velocities of the body
The mean angular velocity during the flexion phase of tailflip 1 was between −6.3×103 and −7.8×103 ° s−1 in small
(11 mm) shrimps compared with between −3.2×103 and
−4.4×103 ° s−1 in large (>60 mm) shrimps. Maximum
instantaneous values attained (over a 5 ms interval) during
flexion were between −1.0×104 and −1.3×104 ° s−1 in small
shrimps and between −6.1×103 and −7.8×103 ° s−1 in large
shrimps.
Flexion rates were linearly related to shrimp total body
length. The mean and maximum angular velocities achieved
during flexion 2 were slower than during flexion 1 (paired t-
tests: t24=5.67, P<0.0001; t24=5.56, P<0.0001 for mean and
maximum angular velocities respectively). Therefore, separate
regression lines were fitted to these data (Fig. 7A,B), all of
which had significant slopes (Table 1).
Mean angular velocity during the re-extension phase of a
tail-flip was slower than that during the flexion phase (paired
t-tests on absolute values: t24=7.41, P<0.0001 for tail-flip 1,
and t19=5.18, P<0.0001 for tail-flip 2). Linear regressions (Fig.
7A,B; Table 1) indicated that both the mean and maximum
angular velocities decreased significantly with body length.
Mean angular velocities during re-extension decreased from
between +2.9×1031 and +4.2×103 ° s−1 in small shrimps to
between +1.2×103 and +2.2×103 ° s−1 in large shrimps, whilst
maximum values decreased from between +4.1×103 and
+9.1×103 ° s−1 to between +3.0×103 and +5.0×103 ° s−1
respectively. The mean angular velocity during re-extension of
tail-flip 1 was slightly greater than that for tail-flip 2 of a
swimming bout (paired t-test: t19=2.18, P=0.042), but there
was no significant difference in the maximum angular
velocities attained (t24=1.63, P=0.12).
Tail-flip swimming in brown shrimps 1779
Maximum angular acceleration of the body angle during the
flexion phase
Maximum angular acceleration values for the flexion phase
of tail-flip 1 changed linearly from approximately −2×105 ° s−2
in small shrimps to approximately −5×104 ° s−2 in large ones
(Fig. 7C; Table 1). There was no significant difference in this
measure between the flexion phases of tail-flip 1 and tail-flip
2 in a swimming bout (paired t-test: t24=0.65, P=0.52).
5×103
0
Distance (mm)
−5×103
−10×103
B
10×103
5×103
Mean velocity (m s−1)
−5×103
−10×103
−15×103
0
Maximum angular
acceleration
(degrees s−2)
A
120
80
40
0
0
−50×10−3
Body length (mm)
10 20 30 40 50 60
0
1.5
C
−150×10−3
−250×10−3
Fig. 7. Body angle measurements during tail-flip 1 (filled symbols)
and tail-flip 2 (open symbols). (A) Mean angular velocity of flexion
phases (negative values) and re-extension phases (positive values).
(B) Maximum instantaneous angular velocity attained during flexion
and re-extension phases. (C) Maximum instantaneous angular
acceleration attained during flexion phases. Fitted regressions are for
the pooled data unless tail-flip 1 and tail-flip 2 values differed
significantly in paired t-tests (see text). Solid line, pooled data or
(when significantly different) tail-flip 1 data. Dashed line, tail-flip 2
data. Fitted lines were derived from the linear regression parameters
given in Table 1.
10
20
30
40
50
60
70
B
1.0
0.5
0
0
70
Maximum velocity
(m s−1)
(degrees s−1)
Maximum angular velocity
Mean velocity of the centre of mass during multiple tail-flips
The mean velocity of the centre of mass measured from
high-speed video recordings agreed very closely with the mean
velocity for escapes in response to predatory cod measured
from conventional video recordings (50 frames s−1), and these
A
(degrees s−1)
Mean angular velocity
Swimming kinematics
Distance travelled per tail-flip
The displacement the centre of mass in tail-flip 1 increased
as a positive function of total body length from 10–30 mm for
small (11 mm) shrimps to 50–120 mm for large (>60 mm)
shrimps. The relationship between total body length and
distance travelled by the centre of mass per tail-flip was
described by a quadratic regression (Fig. 8A) in which both the
positive (value a in Table 1) and negative (value b in Table 1)
slope coefficients were significant (t59=3.62, P<0.001 and
t59=2.01, P<0.05 respectively). Over the size range of shrimps
used in the experiments (11–69 mm), the fitted regression line
predicts a rise in displacement with total body length to a peak
value of 94 mm for a 69 mm shrimp. However, transformed
into body length equivalents, the peak value predicted was
1.8 body lengths for a 28 mm shrimp compared with minimum
values of 1.4 body lengths for both 11 mm and 69 mm shrimps
(the highest measured value was 2.8 body lengths by a 33 mm
shrimp).
The distance travelled by the centre of mass during tail-flip
2 of a swimming bout was not significantly different from that
in tail-flip 1 (paired t-test, t19=1.81, P>0.05), and values for
subsequent tail-flips were also similar. Values have therefore
been pooled in the above regression analysis.
10
20
30
40
50
10
20 30 40 50
Body length (mm)
60
70
C
2.0
1.0
0
0
60
70
Fig. 8. Displacement variables of the shrimp’s centre of mass.
(A) Horizontal displacement per tail-flip. (B) Mean velocity over
entire tail-flip swimming bouts. (C) Maximum velocity attained
during flexion phases. Crosses in B represent recordings at
50 frames s−1 with a natural stimulus (cod); other symbols in A–C are
from high-speed video recordings (200 frames s−1) using an artificial
stimulus. Filled circles, tail-flip 1; open circles, tail-flip 2; squares,
subsequent tail-flips; diamonds, entire swimming bout. Fitted lines
were derived from quadratic regressions (see Table 1).
1780 S. A. ARNOTT, D. M. NEIL AND A. D. ANSELL
data have therefore been pooled. The lowest mean velocity
measured was 0.26 m s−1 by an 8 mm shrimp, and the highest
was 1.42 m s−1 by a 46 mm shrimp. There was a tendency for
values to decline in the largest (>60 mm) shrimps, although
only a few shrimps were filmed within this size range. A
quadratic regression fitted to the data (Fig. 8B; Table 1) had
both positive and negative slope coefficients that were
significant (t62=6.65, P<0.0001 and t62=9.52, P<0.0001
respectively), and had a predicted peak value of 1.07 m s−1 (at
a body length of 52 mm). In terms of body length equivalents,
mean velocity decreased with body length from approximately
37 body lengths s−1
in
the
smallest
shrimps
and
14 body lengths s−1 in the largest.
Maximum velocity of centre of mass during the flexion phase
of the tail-flip
The lowest maximum velocity of the centre of mass for a
single tail-flip was 0.59 m s−1 by a shrimp of 20 mm, and the
highest was 2.31 m s−1 by a shrimp of 57 mm. A quadratic
regression fitted to the data (Fig. 8C; Table 1) had both
positive and negative slope coefficients that were significant
(t71=5.19, P<0.0001 and t71=3.57, P<0.001 respectively) and
had a predicted peak value of 1.8 m s−1 (at a shrimp length of
58 mm). In terms of body length equivalents, maximum
velocity decreased with body length from approximately
60 body lengths s−1
in
the
smallest
shrimps
to
25 body lengths s−1 in the largest.
There was no significant difference between the maximum
velocity achieved by the centre of mass during the first and
second tail-flips of each escape (paired t-test, t24=0.26,
P=0.80). Subsequent tail-flips (up to the fifth of an escape
swimming bout) were also very similar, and so these data have
been pooled in the regression analysis.
Discussion
Thrust-producing mechanisms during tail-flip swimming
In crayfish (Wine and Krasne, 1972; Wine, 1984) and
nephropid lobsters (Newland et al. 1988; Newland and Neil,
1990a), tail-flips mediated by medial giant fibres (in response
to a sudden frontal attack) and by the non-giant circuitry
produce a curling of the tail-fan under the body as flexion is
propagated along the abdomen. This results in little or no
pivoting of the cephalothorax about the centre of mass. In this
‘single oar’ propulsion system, the majority of thrust is
produced by movements of the abdomen through the water and
is due to the inertial forces associated with the acceleration of
the added mass of water (Webb, 1979). These forces are
largely attributable to movements of the tail-fan, which has a
large surface area, is furthest from the point of flexion and has
a high velocity relative to the animal’s centre of mass. Daniel
and Meyhöfer (1989) have also shown that, in the dock shrimp
Pandalus danae, an additional and significant source of thrust
is created by ‘squeeze forces’ towards the end of flexion as
trapped water is ejected from between the abdomen and
cephalothorax.
Tail-flip flexion in Crangon crangon is mainly confined to
the anterior region of the abdomen, causing both it and, to a
lesser degree, the cephalothorax to pivot about the shrimp’s
centre of mass in a ‘jack-knife’ manner. This jack-knife
mechanism, which was common to all forms of tail-flip escape
swimming observed in this study and to shrimps of all lengths,
regardless of stimulus direction, resembles more closely the
flexion pattern of mysid tail-flips (Neil and Ansell, 1995). In
such cases, it would appear that thrust is generated using a
‘twin oar’ system, since thrust may be produced both by the
expanded head-fan and tail-fan. Another contributing factor of
jack-knife tail-flips may derive from greater squeeze forces
(Daniel and Meyhöfer, 1989) produced by the channelling of
water trapped between the abdomen and the cephalothorax.
The jack-knife tail-flip mechanism is promoted by the
lightly armoured chelipeds and carapace of C. crangon,
making the mass of the cephalothorax more equal to that of the
abdomen and bringing the centre of mass close to the point of
flexion. In lobsters and crayfish, the mass of the cephalothorax
and the heavily armoured claws suppresses the jack-knife form
of tail-flip so that it only occurs (mediated by the lateral giant
fibres) in response to a sudden attack from the animal’s rear
(Wine and Krasne, 1972; Neil and Ansell, 1995).
Neuronal control of Crangon crangon tail-flips
The question of the involvement of the giant fibres in C.
crangon tail-flips remains to be addressed. This shrimp
certainly possesses both medial and lateral giant fibres
(Johnson, 1924), and the types of stimuli employed, visual and
vibrational, are known to be sufficient to activate giant fibres
in other crustaceans (Wine, 1984). Furthermore, short response
latencies (10–15 ms) were typical of the escapes observed,
suggesting that the giant fibres were involved, but this must be
interpreted with caution because a more precisely defined
stimulus is needed to measure this parameter specifically.
Unlike the clear differences between lateral- and medialgiant-mediated tail-flips in crayfish and lobsters (Wine and
Krasne, 1972), both frontal and rear attacks to C. crangon
produced fairly similar types of responses. Within the lateral
and vertical escapes produced, there was no obvious difference
in the trajectories between the frontal and rear attacks
(although this was not assessed in detail). Therefore, if the two
giant fibre systems of C. crangon are responsible for conveying
anterior and posterior stimuli to the abdominal flexor motor
system, then their output connections would be expected to
have a greater degree of similarity than in crayfish.
In frontal attacks, there is a clear difference between the
escape trajectories of C. crangon compared with crayfish and
lobsters in that the former do not produce escapes propelled
directly backwards along the substratum. Whilst this may in
part be due to differing electrophysiological connections, the
body posture of C. crangon probably also plays an important
part in excluding this behaviour. Shrimps normally adopt a
resting posture with their entire abdomen in a fully extended
position and flat against the substratum. This causes the initial
thrust produced by both the head-fan and tail-fan at the
Tail-flip swimming in brown shrimps 1781
beginning of an escape to be directed downwards against the
substratum, creating vertical lift.
Several differences were detected between the kinematic
properties of tail-flip 1 and tail-flip 2 of an escape. Tail-flip 1
had a longer flexion and shorter re-extension phase than tailflip 2 and produced higher angular velocities. Some of these
differences are undoubtedly caused by the fact that tail-flip 1
started from a fully extended body position in contact with the
substratum, but some differences may also be attributable to
the involvement of giant fibres in the first flip, with a non-giant
pathway mediating later flips. Confirmation of this requires
direct recordings by electrophysiological methods.
Size-dependent kinematic variability
Daniel and Meyhöfer (1989) calculated that, for a shrimp of
given dimensions and with an isometric growth pattern, there
is a unique body length that maximises the tail-flip kinematic
performance. This arises because of the non-uniform scaling
relationships between the stress limits of the abdominal
muscle, the body dimensions and the translational and
rotational components of tail-flip thrust.
In C. crangon, the abdomen to total length (A:L) ratio and
the cross-sectional area of abdominal fast muscle (F) were both
found to scale isometrically with respect to total length.
Despite these isometric dimensions, wet mass increased at a
rate greater than that expected for isometric growth. A possible
cause for this may be that the shrimp’s more dense exoskeleton
becomes proportionally thicker as length increases, as
suggested by the fact that the underwater wet weight of C.
crangon scales to the power 3.48 (Kils, 1981). This may
decrease the efficiency of tail-flips as shrimps increase in
length.
In their study of Pandalus danae, Daniel and Meyhöfer
(1989) predicted that the duration of the flexion stage of the
tail-flip should increase in an approximately linear manner as
shrimp body length increases (derived from Fig. 9 of Daniel
and Meyhöfer, 1989), and this was found to be true for C.
crangon (Table 1; Fig. 6B). They also predicted an increase in
tail-flip performance with shrimp length until a peak value was
reached (in the case of P. danae, at a body length of 60 mm).
The curves fitted to the mean and maximum velocity values in
C. crangon (Table 1; Fig. 8B,C) clearly indicate that tail-flip
performance increases as the smallest individuals increase in
length, at least between 5 and 50 mm. Some caution is required
in interpreting the maximum velocity values of the smallest
shrimps since, at these sizes, underestimation of the peak
velocity attained may occur. This is because velocity was
measured from only eight images per tail-flip compared with
as many as 24 images per tail-flip in larger shrimps (because
of the size-dependent differences in tail-flip duration). The
potential error arising from this source was tested using a
simulation in which a mathematically derived curve (with the
same shape as the tail-flip velocity data) was repeatedly
‘sampled’ at different rates per cycle, starting from different
points along the curve. From the simulation, peaks were
estimated to lie between 88.7 % and 96.9 % (median 94.7 %)
of their true value (calculated from the curve function) at six
samples per cycle compared with 98.8 % (median 99.5 %) at
24 samples per cycle. In our high-speed video results, the
predicted maximum velocity of the smallest shrimps (11 mm)
was approximately 0.68 m s−1 compared with 1.82 m s−1 for the
larger, fastest shrimps. In the worst case of underestimation,
the value for the smallest shrimps would actually be 0.76 m s−1
(0.71 m s−1 with median error). The error arising from the
lower sampling rate frequencies of small shrimps will therefore
have a negligible effect on the conclusions drawn from the
results.
The fitted mean and maximum velocity curves for C.
crangon (Table 1; Fig. 8B,C) suggest that a peak in tail-flip
performance may occur in shrimps of between 52 and 58 mm.
However, the data presented here provide inconclusive proof
of these peaks because of the small number of shrimps longer
than 60 mm that were filmed and because shrimps approaching
the maximum reported size for the species (80–90 mm; Tiews,
1970) were not available for study.
Orientation of the body whilst tail-flipping
In the majority of tail-flips analysed, C. crangon performed
a pronounced roll about its longitudinal axis during the first
flexion of an escape response. If this did not occur, and the
shrimp escaped with an initial vertical trajectory, it usually
rolled onto its side during the first re-extension phase. The
division between these two types of behaviours is not absolute
since different degrees of initial roll were observed. In both
cases, however, shrimps then swam on their side in a horizontal
direction during subsequent tail-flips before righting
themselves again as they settled back onto the substratum.
Incidents in which shrimps did not escape on their side at all
(i.e. when they swam in a head-down, tail-up position) were
rare. This strong tendency for C. crangon to swim on their side
contrasts with the typical tail-flip behaviour of many decapod
crustaceans (e.g. Wine and Krasne, 1972; Webb, 1979; Sillar
and Heitler, 1985; Jacklyn and Ritz, 1986; Wilson and Paul,
1987; Spanier et al. 1991; Newland et al. 1992a), which
generally tail-flip in an upright position and have dynamic selfrighting mechanisms that maintain this orientation during an
escape (Newland and Neil, 1990b; Newland et al. 1992b).
Body roll during the first tail-flip of an escape in C. crangon
is fundamental in re-directing the vertical thrust from jackknife tail-flips to produce a horizontal escape trajectory.
Without it, shrimps would continue to tail-flip vertically (if no
somersault is performed), translating them away from the
refuge provided by the sediment, against which they are cryptic
and within which they are able to hide by quickly burying
themselves once they resettle (Pinn and Ansell, 1993).
Escaping too far off the substratum may also be
disadvantageous because the silhouette image of the shrimp
created against the water surface will make it conspicuous to
visual predators (Thetmeyer and Kils, 1995). A further
advantage offered by the body roll is that it can occur to the
shrimp’s left or right in an unpredictable (‘protean’) manner,
1782 S. A. ARNOTT, D. M. NEIL AND A. D. ANSELL
a factor that may assist in evading an attacking predator (Driver
and Humphries, 1988).
An initial roll towards one side of the body has also been
shown to take place in mysid shrimps (Kaiser et al. 1992; Neil
and Ansell, 1995) which, as described above, also use jackknife tail-flips and form an expanded head-fan. The relatively
small size of C. crangon (and mysids) and their comparatively
thin (i.e. light) exoskeleton are probably important features that
enable them to adopt these characteristics. Larger crustaceans
with heavily calcified exoskeletons have to generate a greater
proportion of vertical thrust when tail-flipping in order to
maintain height above the substratum. This is facilitated by
being in an upright position because the rotational forces
generated during abdominal flexion (Daniel and Meyhöfer,
1989) create vertically directed thrust and, in the case of
heavily calcified scyllarid lobsters, because their large antennal
scales act as ‘ailerons’ (Jacklyn and Ritz, 1986).
A jack-knife tail-flip mechanism and swimming on one side
of the body seem to be incompatible with anachoresis (i.e.
living within crevices or burrows) since tail-flip movements
would become obstructed by the walls of a narrow
passageway. The use of jack-knife tail-flipping, head-fan
formation and swimming on one side therefore appear to be
intrinsically linked adaptations adopted by small crustaceans
living in relatively unconfined habitats.
Steering of tail-flips in the animal’s roll plane
In laterally directed escapes, the sideways roll was often
evident within the first few frames in which movement was
detected (Fig. 5). It is not obvious how these rotational forces
were brought about. Neil and Ansell (1995) noted that, when
the mysid Praunus flexuosus rolled on to its side during its first
tail-flip, the antennal scales and uropods were expanded
asymmetrically and acted as rotors which contributed towards
the forces bringing about the body roll. Occasionally,
asymmetrical spreading of the antennal scales or uropods was
observed in C. crangon, but this was not noticeable in many
of the escapes and was not necessary for body roll to occur.
An alternative may be that the roll is brought about by
asymmetrical muscle activity in the shrimp’s abdomen.
Newland and Neil (1990b) have shown that, during tail-flip
swimming in Nephrops norvegicus, dynamic righting reactions
in the animal’s roll plane are brought about primarily by
rotation of the abdomen relative to the cephalothorax. C.
crangon possess a set of oblique fast muscles spanning this
joint (S. A. Arnott, personal observations), and it is possible
that these serve to tilt the shrimp away from the upright rather
than towards it at the beginning of an escape.
Another possible contributor to roll is suggested by the
observation that, in the palinurid lobster Jasus lalandii,
asymmetrical movements of the swimmerets can cause
movements in the animal’s roll plane during tail-flips (Cattaert
et al. 1988). In the video sequences of C. crangon, the pleopods
were obscured from the camera’s view by the abdomen during
the initial stages of an escape. However, when shrimps
performed an initial vertical flexion and then rolled onto their
side during the re-extension phase, pleopod movement was
sometimes visible. This suggests that the pleopods may assist
in bringing about body roll, at least under some circumstances.
In scyllarid lobsters, roll manoeuvres are controlled during
the glide phase of tail-flips (i.e. at the end of the flexion phase),
once the animal is swimming within the water column, by
asymmetrically raising or lowering their antennal scales
(Jacklyn and Ritz, 1986). Whether C. crangon are able to use
a similar form of steering during subsequent tail-flips is
uncertain.
Steering in the animal’s pitch plane
When C. crangon do perform a body roll during their first
tail-flip and then swim on their side, control of rotation in the
shrimp’s pitch plane brings about horizontally directed
steering. During the largest steering manoeuvres in the rostral
direction, a backward beat of the pleopods was often
observed during the preceding re-extension phase (Fig. 5).
This may serve to prevent the cephalothorax from pivoting
about the centre of mass as the abdomen re-extends, thereby
adjusting the shrimp’s orientation in preparation for the next
flexion phase and directing it along a new trajectory.
However, large pitching manoeuvres also occurred without
the assistance of pleopod activity. In these cases, spatial and
temporal adjustments in muscle activation patterns within the
abdomen may have contributed to the pitching movements,
as occurs in larger decapods (Wine, 1984; Newland and Neil,
1990a). Additionally, the head-fan was not always fully
retracted during the re-extension phase of a rostrally directed
pitch movement, and this may also have affected the
shrimp’s pitch orientation. At present, tail-flip steering in
C. crangon remains poorly understood and warrants further
investigation.
Comparison of swimming kinematics
Neil and Ansell (1995) compared the tail-flip performances
reported in the literature for six species of decapods, two
species of mysids and one species of euphausid across a range
of body lengths covering 10–270 mm. Maximum velocities
ranged from 0.6 m s−1 for the nephropid lobster Nephrops
norvegicus (Newland et al. 1988) to 2.8 m s−1 for the caridean
shrimp Pandalus danae (Daniel and Meyhöfer, 1989), with
most species having peak velocities of less than 1 m s−1.
Among these species, Crangon crangon ranks as a relatively
fast tail-flip swimmer according to the maximum velocity of
1.8 m s−1 for a 58 mm shrimp predicted by the regression
equation in Table 1, but it is difficult to make accurate
comparisons because of the different frame rates and
conditions used in the various studies.
In comparison with one of the common predators of C.
crangon, the cod Gadus morhua (Tiews, 1970; Pihl, 1982;
Berghahn, 1996), the results presented here suggest that
shrimps will not be able to escape a pursuing cod by using
speed alone. Juvenile cod are able to consume shrimps up to
36 % of their own body length (Arnott, 1996). According to
Wardle (1975), a cod with a length of 100 mm can achieve a
Tail-flip swimming in brown shrimps 1783
maximum velocity of up to 1.7 m s−1, which is faster than
shrimps in the edible range 10–36 mm (0.6–1.6 m s−1). More
significant is the fact that the maximum velocity of cod
continues to increase with length so that, for a cod of 300 mm,
Wardle (1975) predicts a value of 3.3 m s−1. This is
considerably faster than C. crangon of any length.
Furthermore, because the tail-flip velocity of C. crangon may
level off in shrimps larger that 50 mm, one might expect the
escape success of a large shrimp being attacked by a large cod
to be lower than for a small shrimp being attacked by a
proportionally smaller-sized cod. The inability of C. crangon
to out-compete predators in a ‘straight-line race’ suggests that
other strategies such as manoeuvrability (Howland, 1974;
Webb, 1976; Weihs and Webb, 1984), unpredictability (Driver
and Humphries, 1988; Domenici and Blake, 1993) and hiding
on or within the substratum will also be important in
determining the outcome of encounters with predators.
We are grateful to Dr R. S. Batty for generous provision of
and assistance with the high-speed video equipment and to
other staff and students at the Dunstaffnage Marine
Laboratory (DML) for their support with the project. We also
thank Dr B. Bergström and two anonymous referees for their
suggestions on improving the manuscript. Measurements used
for calculating the abdomen length to total length ratio of
shrimps were kindly provided by Håkan Wennhage and Sven
Nilsson. The project was funded by a Natural Environment
Research Council CASE studentship (S.A.A.) between the
University of Glasgow and DML.
References
AL-ADHUB, A. H. Y. AND NAYLOR, E. (1975). Emergence rhythms and
tidal migrations of the brown shrimp, Crangon crangon (L.). J.
mar. biol. Ass. U.K. 55, 801–810.
ARNOTT, S. A. (1996). The tail-flip escape response of the brown
shrimp, Crangon crangon (L.) in the context of predator–prey
interactions. PhD thesis, University of Glasgow, Scotland, UK.
BELLMAN, K. L. AND KRASNE, F. B. (1983). Adaptive complexity of
interactions between feeding and escape in crayfish. Science 221,
779–781.
BERGHAHN, R. (1996). Episodic mass invasions of juvenile gadoids
into the Wadden Sea and their consequences for the population
dynamics of brown shrimp (Crangon crangon). Mar. Ecol. 17,
251–260.
BERGHAHN, R., WIESE, K. AND LÜDEMANN, K. (1995). Physical and
physiological-aspects of gear efficiency in North Sea brown shrimp
fisheries. Helgolander wiss. Meererunters. mar. Invest. 49,
507–518.
CATTAERT, D., CLARAC, F. AND NEIL, D. M. (1988). Anatomical and
physiological organisation of the swimmeret system of the spiny
lobster Jasus lalandii as adaptive components of the tail flick. J.
comp. Physiol. A 162, 187–200.
DANIEL, T. L. AND MEYHÖFER, E. (1989). Size limits in escape
locomotion of carridean shrimp. J. exp. Biol. 143, 245–265.
DOMENICI, P. AND BLAKE, R. W. (1993). Escape trajectories in
angelfish (Pterophyllum eimekei). J. exp. Biol. 177, 253–272.
DRIVER, P. M. AND HUMPHRIES, D. A. (1988). Protean Behaviour: The
Biology of Unpredictability. Oxford: Oxford University Press.
EDWARDS, D. H. (1995). The neural mechanisms of social dominance
status in crustaceans. In Nervous Systems and Behaviour:
Proceedings of the Fourth International Congress of
Neuroethology (ed. M. Burrows, T. Matheson, P. L. Newland and
H. Schuppe), p. 3. Stuttgart: Thieme Medical Publishers. Inc.
HARPER, D. G. AND BLAKE, R. W. (1989). On the error involved in
high-speed film when used to evaluate maximum acceleration of
fish. Can. J. Zool. 67, 1929–1936.
HOWLAND, H. C. (1974). Optimal strategies for predator avoidance:
the relative importance of speed and manoeuvrability. J. theor.
Biol. 47, 333–350.
JACKLYN, P. M. AND RITZ, D. A. (1986). Hydrodynamics of swimming
in scyllarid lobsters. J. exp. mar. Biol. Ecol. 101, 85–99.
JOHNSON, G. E. (1924). Giant nerve fibres in crustaceans, with special
reference to Cambarus and Palaemonetes. J. comp. Neurol. 36,
323–373.
KAISER, M. J., GIBSON, R. N. AND HUGHES, R. N. (1992). The effect
of prey type on the predatory behaviour of the fifteen-spined
stickleback, Spinachia spinachia (L.). Anim. Behav. 43, 147–156.
KILS, U. (1982). The swimming behaviour, swimming performance
and energy balance of the Antarctic krill, Euphausia superba.
Biomass Sci. Ser. 3, 1–121.
KRASNE, F. AND WINE, J. J. (1975). Extrinsic modulation of crayfish
escape behaviour. J. exp. Biol. 63, 433–450.
KRASNE, F. AND WINE, J. J. (1988). Evasion responses of the crayfish.
In Aims and Methods in Neuroethology (ed. D. M. Guthrie), pp.
10–45. Manchester: Manchester University Press.
LANG, F., GOVIND, C. K., COSTELLO, W. J. AND GREENE, S. I. (1977).
Developmental neuroethology: changes in escape and defensive
behavior during growth of the lobster. Science 197, 682–685.
NEIL, D. M. AND ANSELL, A. D. (1995). The orientation of tail-flip
escape swimming in decapod and mysid crustaceans. J. mar. biol.
Ass. U.K. 75, 55–70.
NEWLAND, P. L., CATTAERT, D., NEIL, D. M. AND CLARAC, F. (1992a).
Steering reactions as adaptive components of the tail-flip in the
spiny lobster Jasus lalandii. J. exp. Biol. 164, 261–282.
NEWLAND, P. L., CHAPMAN, C. J. AND NEIL, D. M. (1988). Swimming
performance and endurance of the Norway lobster, Nephrops
norvegicus. Mar. Biol. 98, 345–350.
NEWLAND, P. L. AND NEIL, D. M. (1990a). The tail-flip of the Norway
lobster, Nephrops norvegicus. I. Giant fibre activation in relation to
swimming trajectories. J. comp. Physiol. A 166, 517–527.
NEWLAND, P. L. AND NEIL, D. M. (1990b). The tail-flip of the Norway
lobster, Nephrops norvegicus. II. Dynamic righting reactions
induced by body tilt. J. comp. Physiol. A 166, 529–536.
NEWLAND, P. L., NEIL, D. M. AND CHAPMAN, C. J. (1992b). Escape
swimming in the Norway lobster, Nephrops norvegicus. J. Crust.
Biol. 12, 342–353.
PAUL, D. H. (1990). Neural phylogeny – its use in studying neural
circuits. In Frontiers in Crustacean Neurobiology (ed. K. Wiese,
W.-D. Krenz, J. Tautz, H. Reichert and B. Mulloney), pp. 537–546.
Basel: Birkhäuser Verlag.
PIHL, L. (1982). Food intake of young cod and flounder in a shallow
bay on the Swedish west coast. Neth. J. Sea Res. 15, 419–432.
PINN, E. H. AND ANSELL, A. D. (1993). The effect of particle size on
the burying behaviour of the brown shrimp, Crangon crangon. J.
mar. biol. Ass. U.K. 73, 365–377.
SILLAR, K. T. AND HEITLER, W. J. (1985). The neural basis of escape
swimming behaviour in the squat lobster Galathea strigosa. I.
1784 S. A. ARNOTT, D. M. NEIL AND A. D. ANSELL
Absence of cord giant axons and anatomy of motor neurones
involved in swimming. J. exp. Biol. 117, 251–269.
SPANIER, E., WEIHS, D. AND ALMOG-SHTAYER, G. (1991). Swimming
of the Mediterranean slipper lobster. J. exp. mar. Biol. Ecol. 145,
15–31.
TALLMARK, B. AND EVANS, S. (1986). Substrate related differences in
antipredator behaviour of two gobiid fish species and the brown
shrimp and their adaptive value. Mar. Ecol. Prog. Ser. 29, 217–222.
THETMEYER, H. AND KILS, U. (1995). To see and not be seen: the
visibility of predator and prey with respect to feeding behaviour.
Mar. Ecol. Prog. Ser. 126, 1–8.
TIEWS, K. (1970). Synopsis of biological data on the common shrimp
Crangon crangon (Linnaeus, 1758). FAO Fisheries Report. 57,
1167–1224.
WARDLE, C. S. (1975). Limit of fish swimming speed. Nature 255,
725–727.
WEBB, P. W. (1976). The effect of size on the fast-start performance
of rainbow trout Salmo gairdneri and a consideration of piscivorous
predator–prey interactions. J. exp. Biol. 65, 157–177.
WEBB, P. W. (1979). Mechanics of escape responses in crayfish
(Orconectes virilis). J. exp. Biol. 79, 245–263.
WEIHS, D. AND WEBB, P. W. (1984). Optimal avoidance tactics in
predator–prey interactions. J. theor. Biol. 106, 189–206.
WILSON, L. J. AND PAUL, D. H. (1987). Tailflipping of Munida
quadrispina (Galatheidae): conservation of behavior and
underlying musculature with loss of anterior contralateral flexor
motoneurones and motor giant. J. comp. Physiol. A 161,
881–890.
WINE, J. J. (1984). The structural basis of an innate behaviour pattern.
J. exp. Biol. 112, 283–319.
WINE, J. J. AND KRASNE, F. B. (1972). The organisation of escape
behaviour in the crayfish. J. exp. Biol. 56, 1–18.
ZAR, J. H. (1996). Biostatistical Analysis. Upper Saddle River, NJ:
Prentice Hall Inc.